Business judgment on AI products
There is public usage data. The useful view is what it does, its scale, and whether the change lasts.
01
Start inside the user's day · Public facts + observable behavior · 2026-08-28
秘塔 has no ads and answers with sources attached. Monthly visits: 4.5 million.
Public materials do not yet show how users complete this job today or what they replace.
The product targets friction in this job, but public user evidence does not yet show the cost, frequency, or consequence of leaving it unsolved.
A public record shows 1.38M mau and -1.62% month-over-month growth. That explains the attention, but product-level retention and payment are not yet verified.
① Whether web visits stop declining — the lifeline for independent AI search; under giant siege; three straight months of decline warrants a re-rating; ② Whether "research mode" or search itself starts charging — the COO left the; door open; this is the test of whether search can make money on its o…
Worth dissecting. A public record shows 1.38M mau and -1.62% month-over-month growth. That explains the attention, but product-level retention and payment are not yet verified.
build the brand on a stance giants cannot take — "no ads" attacks the competitor's business model head-on, because the giants dare not give up advertising. Then feed the company with a separate niche product. Separating the "business card product" from the "granary product" makes you far more resilient than a single-product company. "a tool's price ≈ 1/10 of the work value it replaces" is a transferable pricing principle — first calculate what the product saves the user, then derive the price backward, instead of pricing by cost or competitors.
Search itself is free, no paid version today; Revenue comes from Writing Cat (13M+ registered users, ~3-4% paying, mostly; civil servants and students), legal translation, and MetaLaw enterprise; subscriptions; COO Wang Yiwei: "a fair price… ① Whether web visits stop declining — the lifeline for independent AI search; under giant siege; three straight months of decline warrants a re-rating; ② Whether "research mode" or search itself starts charging — the COO left the; door open; this is the test of whether search can make money on its o…
秘塔 has no ads and answers with sources attached. Monthly visits: 4.5 million.
A public record shows 1.38M mau and -1.62% month-over-month growth. That explains the attention, but product-level retention and payment are not yet verified.
An English validation note will follow from the public evidence.
The product claims to help users complete: “秘塔 has no ads and answers with sources attached. Monthly visits: 4.5 million.”. User evidence has not yet verified pain intensity or the cost of doing without it.
The assessment is recorded; an English explanation is pending.
The assessment is recorded; an English explanation is pending.
The assessment is recorded; an English explanation is pending.
02
Market comparison
The Chinese–English market comparison is not complete yet. A conclusion follows only after its coverage and verifiable evidence are recorded.
03
The call and next move come first; the full read retains the evidence and counterevidence.
No ads, straight to the answer AI search — turning "a pile of links you filter yourself" into "one sourced answer you read," then converting that traffic into money through its writing and vertical tools.
Shanghai Metaso Network Technology, founded April 2018, focused on legal AI and writing tools. Founder/CEO Min Kerui (Fudan CS, Oxford math master's; ex-CTO of Bosen Data, ex-chief scientist of Cheetah Mobile); COO Wang Yiwei (Peking Law, LLM at UCL and Columbia; former lawyer at international firms). In-house model MetaLLM (~10B). Team of ~50. Closed a RMB 100M+ round in August 2024 led by Ant Group at a $150M valuation. The company is break-even.
Read: Metaso is a company that started in legal tech and accidentally produced a general-purpose search. Its shrewdness is not out-spending giants but selling the one thing giants cannot give: "no ads."
What it deliberately does not do: no super-app ambitions, no burn-heavy advertising. The product matrix stays in one niche: text and knowledge work.
Searching Baidu or Bing used to mean wading through ads, SEO farms, and portal content — mixed-quality results to filter yourself. "No ads, straight to the answer" cuts that layer out entirely, replacing the link list with a cited, structured answer.
For academic users, the "academic" channel replaces the Google Scholar / CNKI retrieval flow. For writers, research mode compresses "gather → read → write" into "ask → read the answer → process." The core thing it replaces is the old act of "filtering information yourself amid ads and noise."
Read: Metaso is "search as a business card, writing as the granary." Free search builds the brand; vertical paid tools feed the company. Search staying free is deliberate — its job is to be seen, the money comes from other products.
| Dimension | Call |
|---|---|
| Founder-product fit | High — legal-plus-engineering background, real understanding of knowledge work |
| Product insight | "No ads" is the asset giants cannot abandon; a precise and durable cut |
| Execution quality | An in-house 10B model is sufficient for writing and retrieval, cost-priority |
| Timing | In the AI-search war, independent search keeps getting squeezed by big-platform entry points |
Metaso is the healthiest independent AI search company, but the search pool is shrinking. It grew without burning money — via Bilibili content (the COO's personal account has 600K+ followers), word of mouth, and the "no ads" stance — and reached break-even, which very few AI startups can say.
The traffic data must be faced honestly: from a peak of 15M monthly visits in 2024 down to ~4.5M today, independent search space is steadily contracting under siege from big-platform AI search (Doubao, Yuanbao, Quark, and Kimi all ship search). -0.31% and -1.62% MoM mark a plateau, not a cliff, but the direction deserves attention.
Its real value is not search, it is the vertical tools. Writing Cat delivers ~80% of revenue and is profitable; MetaLaw has real lawyer users. Search's value is brand and acquisition, not revenue — a structure that actually protects the company under giant pressure: if search dies, the company survives.
The Ant stake is worth watching: Ant's strategy is "professional agents," and Metaso's writing and legal capabilities fit financial and legal scenarios. If Ant supplies distribution (e.g., Alipay entry), Metaso may gain distribution rare for an independent search company; if the cooperation stays financial, Metaso fights big-platform traffic suppression alone.
① Whether web visits stop declining — the lifeline for independent AI search under giant siege; three straight months of decline warrants a re-rating ② Whether "research mode" or search itself starts charging — the COO left the door open; this is the test of whether search can make money on its own ③ Whether Ant delivers actual distribution (e.g., an Alipay entry) — yes, Metaso gets distribution; no, it stays small and beautiful
Product logic: build the brand on a stance giants cannot take — "no ads" attacks the competitor's business model head-on, because the giants dare not give up advertising. Then feed the company with a separate niche product. Separating the "business card product" from the "granary product" makes you far more resilient than a single-product company.
Pricing structure: "a tool's price ≈ 1/10 of the work value it replaces" is a transferable pricing principle — first calculate what the product saves the user, then derive the price backward, instead of pricing by cost or competitors.
Worth watching, with an unproven streak. The company is real, profitable, and clearly positioned — the independent search worth tracking most closely. But the search pool keeps contracting, and there is no second growth curve beyond Writing Cat. The next hinges on traffic stabilizing and Ant delivering distribution; if neither happens, it stays in its small-and-beautiful corner.
05
Use these searches when the official site is missing or the current link is only a lead.